Bright Light Treatment of Winter Depression a Placebo-Controlled Trial

ORIGINAL ARTICLE Bright Light Treatment of Winter Depression A Placebo-Controlled Trial

Charmane I. Eastman, PhD; Michael A. Young, PhD; Louis F. Fogg, PhD; Liwen Liu, PhD; Patricia M. Meaden, PhD

Background: Bright light therapy is the recommended Results: There were no differences among the 3 groups treatment for winter seasonal affective disorder (SAD). in expectation ratings or mean depression scores after 4 However, the studies with the best placebo controls have weeksoftreatment.However,strictresponsecriteriarevealed not been able to demonstrate that light treatment has a ben- statistically significant differences; after 3 weeks of treat- efit beyond its placebo effect. ment morning light produced more of the complete or al- most complete remissions than placebo. By 1 criterion (24- Methods: Ninety-six patients with SAD completed the item SIGH-SAD score Ͻ50% of baseline and Յ8), 61% of study. Patients were randomly assigned to 1 of 3 treat- the patients responded to morning light, 50% to evening ments for 4 weeks, each 1.5 hours per day: morning light light, and 32% to placebo after 4 weeks of treatment. (average start time about 6 AM), evening light (average start about 9 PM), or morning placebo (average start about Conclusions: Bright light therapy had a specific antidepres- 6 AM). The bright light (Ϸ6000 lux) was produced by sant effect beyond its placebo effect, but it took at least 3 light boxes, and the placebos were sham negative-ion gen- weeks for a significant effect to develop. The benefit of light erators. Depression ratings using the Structured Inter- over placebo was in producing more of the full remissions. view Guide for the Hamilton Depression Rating Scale, SAD version (SIGH-SAD) were performed weekly. Arch Gen Psychiatry. 1998;55:883-889

HE SHORTER photoperiodand bright light could have been entirely due to decreaseinsunlightexposure placebo effects. Another strategy has been experienced by people living to use evening light as the placebo control in temperate and higher lati- for morning light.14 However, some inves- tudes during the winter1-3 is tigators have shown that evening light hypothesized to be the main trigger for win- can be as antidepressant as morning T 15 ter depression or seasonal affective disorder light, limiting the usefulness of evening (SAD).4 Supplementation with bright arti- light as an inert placebo. ficial light is therefore logical and is also the medically recommended treatment of See also pages 861, choice.5,6 There is no doubt that bright light 863, 875, and 890 therapy can reduce and even eliminate the symptoms of winter SAD. However, it has re- A recent study successfully imple- mained difficult to demonstrate that bright mented an inert placebo control, a light box light has an antidepressant effect beyond that emitted no visible light, but had a pi- its placebo effect. One reason stems from lot light and a hum when turned “on.”16 Pa- the problems inherent in finding an appro- tients were told that the purpose of the ex- priateplacebocontrolforlightboxtreatment. periment was to study infrared, nonvisible Many researchers used dim light as the light. The deception was successful; there placebo control for bright light. However, was no difference in expectations between From the Biological Rhythms even in the earliest study,7 most patients the patients assigned to bright light and Research Laboratory, predicted that bright light would be more those assigned to placebo. Unfortunately, Departments of Psychology (Drs Eastman and Liu) and helpful. Because expectations for improve- there was also no difference in antidepres- Psychiatry (Drs Young, ment are believed to account for a large com- sant response. 8-12 Fogg, and Meaden), ponent of placebo effects and placebo ef- Several years ago, we used an inert Rush-Presbyterian-St Luke’s fects account for a large component of (deactivated) negative ion generator as the Medical Center, Chicago, Ill. antidepressant effects,13 the superiority of placebo control for bright light treat-

ARCH GEN PSYCHIATRY/ VOL 55, OCT 1998 883

©1998 American Medical Association. All rights reserved. Downloaded From: https://jamanetwork.com/ on 09/26/2021 PATIENTS AND METHODS morning and evening treatments, if it existed. Bedtime fell within a 1-hour window so that patients were allowed 7 to 8 PATIENTS hours in bed. Naps were permitted within a 6-hour window in the middle of the waking period. Compliance with the sleep Patients were recruited through advertisements and the lo- schedule was encouraged and monitored by requiring that cal media. They were diagnosed by the usual criteria for SAD,7 patients call a time-stamp answering machine or voice mail but in addition we required the “atypical” symptoms of 3 times each morning, during the 1.5 hours after waking, and increased appetite or weight and increased sleep. Most SAD by reviewing daily sleep logs at least once per week. Patients patients have these symptoms, and this provided a more ho- were not permitted to drink alcohol during the 5 weeks or mogeneous and more typical sample. Patients also had to score to drink caffeine in the 6 hours before their earliest bedtime. 21 or more on the first 24 items of the Structured Interview Patients went to 2 different buildings during weekly vis- Guide for the Hamilton Depression Rating Scale, SAD Ver- its to the medical center. The clinical staff in the Depart- sion (SIGH-SAD),18 ie, the original 17-item Hamilton De- ment of Psychiatry, Rush-Presbyterian-St Luke’s Medical Cen- pression Scale plus 7 atypical items, the 24 items that were ter, Chicago, Ill, performed weekly SIGH-SAD ratings, and intended to rate severity. All patients were free of psycho- were blind to the patients’ type of equipment and time of day tropic medications for several months and none had previ- of treatment. The research assistants in the Biological Rhythms ously tried bright light or negative ion treatment. None had Research Laboratory dispensed equipment, collected ques- any complicating medical conditions based on medical his- tionnaires, and monitored compliance to the protocol. They tories, physical examinations, and blood and urine evalua- were not blind to type of equipment or time of day of treat- tions. Patient characteristics are listed in Table 1. ment, but they did not know which patients received active generators and which received placebos. This division of func- EQUIPMENT tions between the 2 locations helped to maintain the blinding of the SIGH-SAD raters. The ion generators were shiny black cylinders, 32 cm in Treatment was administered 6 days per week; it was omit- height, 15 cm in diameter, with 3 small lights on the front ted on the weekly visit day. Patients started the protocol in that changed rapidly between red and green. Because the January or February, except for 3 who started in November. negative ion mechanism made a soft hissing sound, a white- Before the study began patients were given a packet of noise generator was added to active and placebo genera- information from the scientific and lay literature promoting tors. The patient consent forms explained that they might negative ion and light treatment of SAD. At the end of the receive active or deactivated generators, but that they would baseline week, after patients had known their treatment as- not be able to tell which type they had received. Two gen- signment for 1 week, they completed an expectation ques- erators were set up on a desk or table in front of the pa- tionnaire that asked them to rate, on a 7-point scale, how they tient, 38.1 cm apart, with 38.1 cm between the patient and expected to feel at the end of the 4-week treatment. Possible each generator. answers ranged from “will feel as good as summer” (1) to “no The light box was 65 cm wide and 43.5 cm tall (Apollo change” (7). They also rated how confident they were in that Light Systems, Orem, Utah). It contained 6 horizontally guess, from “very confident” (1) to “not at all confident” (7). mounted cool-white fluorescent lamps. Patients sat at a table At the end of each week patients completed the 21- or desk with the light box directly in front of them at a dis- item Beck Depression Inventory,19 to which we added 4 tance of 38.1 cm. They usually read during the treatment atypical items. The 4 items covered increases in appetite, time, which produced an illuminance of about 6000 lux. weight, sleep, and sleepiness. The study protocol was approved by the institutional PROCEDURES review board and all patients gave written informed consent before the study began. The 5-week protocol consisted of a baseline week followed by 4 consecutive treatment weeks. Patients were told DATA ANALYSIS that they would be randomly assigned to 1 of 4 treatments: morning generator, evening generator, morning light, We analyzed data from patients assigned to morning light or evening light. All treatments were 1.5 hours in dura- (ML), morning placebo (MP), and evening light (EL). Pa- tion. The morning treatments were taken as soon as pos- tients and raters did not know that very few patients were sible after waking and the evening treatments were before assigned to active ion generators and none were assigned bed, with 1 hour maximum between the end of treatment to evening generators. Patients were randomly assigned to and bedtime. the 3 major groups with balancing for sex. Statistical analy- Sleep schedules were negotiated with each patient, and ses included repeated-measures analyses of variance (ANO- were followed during the 5-week study, including week- VAs) and log-linear analyses20 for depression scores, and ends. The wake-up time had to be earlier than usual, to make 1-way ANOVAs and Tukey honestly significant difference time for the morning treatments. Even patients assigned post-hoc tests for expectation ratings and sleep param- to evening treatments had to wake up earlier because they eters. In all statistical tests we used a significance level of were asked to stay indoors for the first 1.5 hours after .05, 2 tailed, unless otherwise specified. Summary statis- waking. This was to help us find a difference between tics are presented as mean ± SD.

ment.17 To make ion treatment plausible, patients were with 1 hour of morning bright light and 1 hour of morn- given an information packet including the statement that, ing placebo in a balanced crossover design. Expectation in general, people are exposed to more negative ions in ratings showed that the patients expected their light box the summer as well as to more light. Patients were treated to make them feel slightly better than their generator. De-

ARCH GEN PSYCHIATRY/ VOL 55, OCT 1998 884

©1998 American Medical Association. All rights reserved. Downloaded From: https://jamanetwork.com/ on 09/26/2021 Table 1. Demographic and Table 2. Sleep Parameters* Clinical Characteristics of Patients* Morning Morning Evening Morning Morning Evening Light Placebo Light Characteristic Light Placebo Light Scheduled time† No. of patients (M/F) 33 (5/28) 31 (3/28) 32 (5/27) Wake-up 6.1 (1.3) 6.2 (1.1) 6.4 (0.9) Age, y 35.5 (10.7) 37.0 (9.2) 37.7 (11.3) Latest bedtime‡ 23.1 23.2 23.4 Race/ethnicity, Wake-up† No. of patients Pre-baseline week 7.9 (1.4) 7.9 (1.2) 7.6 (1.1) White 28 28 31 Baseline week 6.2 (1.3) 6.1 (1.2) 6.5 (1.0) Black 4 0 1 Treatment weeks§ 6.1 (1.3) 6.1 (1.2) 6.5 (0.9) Asian 0 1 0 Sleep onset† Hispanic 1 2 0 Pre-baseline week 23.9 (1.6) 23.8 (1.1) 23.5 (1.1) % Married 57.6 45.2 34.4 Baseline week 23.3 (1.4) 23.4 (1.1) 23.5 (1.0) Education, y 16.2 (2.4) 16.1 (2.5) 16.7 (2.4) Treatment weeks§ 23.1 (1.3) 23.3 (1.1) 23.7 (1.0) Duration of SAD, y 14.3 (7.9) 13.1 (6.8) 14.8 (8.0) Nighttime sleep࿣ Age at onset of SAD, y 22.1 (9.8) 22.8 (7.7) 21.7 (10.1) Pre-baseline week 7.8 (0.9) 8.0 (0.8) 7.9 (1.2) Previous treatment, % Baseline week 6.7 (0.6) 6.6 (0.5) 6.8 (0.5) Medication 36.4 29.0 34.4 Treatment weeks§ 6.9# (0.4) 6.7 (0.5) 6.6 (0.4) Psychotherapy 51.5 48.4 65.6 Total sleep time¶ Pre-baseline week 8.3 (0.9) 8.3 (1.0) 8.1 (1.2) *Data are given as mean (SD) unless otherwise indicated. SAD indicates Baseline week 7.0 (0.7) 6.9 (0.6) 7.0 (0.7) seasonal affective disorder. Treatment weeks§ 7.2# (0.5) 7.0 (0.6) 6.9 (0.5)

*All data are given as mean (SD). spite this slight advantage for light, the benefit of light †Hours past midnight. ‡Seven hours before scheduled wake-up time. treatment over placebo was not statistically significant. §Mean of all 4 treatment weeks. For the current study, we improved the design to ࿣Time (in hours) from sleep onset to wake-up minus any awakenings of increase the chances of finding a significant effect of bright 10 minutes or more. light treatment, should it exist. We increased the daily ¶Nighttime sleep plus naps. #Significantly different from evening light (PϽ.05). duration of light treatment from 1.0 to 1.5 hours, and increased the weeks of light treatment from 2 to 4. We coached the staff to be enthusiastic and positive about both treatments, as opposed to the rather neutral and pes- pectation vs SIGH-SAD difference from pre-baseline to simistic attitude displayed in our previous study. We also treatment; at treatment week 3, r = − 0.18, P = .04; at treat- changed to a simpler parallel design, instead of a cross- ment week 4, r = − 0.14, P = .09; both with df = 94, 1- over design, and increased the sample size. An evening tailed tests). light group was included to address the controversy about whether time of day of treatment is important. We also SLEEP wanted to increase the expectations for the generator treatment to be more similar to those for the light treatment. Sleep logs were analyzed to determine whether there Therefore, we had more impressive ion generators built were differences among groups (Table 2). If there and to increase the “dose” we gave each patient 2 units were differences in sleep schedules or other sleep instead of 1. parameters between light and placebo treatments, then these nonspecific factors might contribute to any differ- RESULTS ences in antidepressant response. One-way ANOVAs were used to compare treatment groups. Within-group During the course of 6 winters, 96 patients completed comparisons are only descriptive. There were no statis- the study (Table 1) and 25 dropped out during the 5- tically significant differences among groups in the week protocol (ML = 8, EL = 8, MP = 9). scheduled wake-up times, which averaged about 6 AM. Actual wake-up time was about an hour or 2 earlier EXPECTATIONS during the study (baseline and treatment) than before the study (pre-baseline), when there were no restric- The expectation ratings showed that all groups ex- tions on sleep. However, sleep onset times did not pected their treatment to be very successful in reducing change as much as wake-up times. Consequently, their depression (ML = 3.6 ± 1.1, MP = 3.7 ± 1.2, nighttime sleep was about an hour less during the EL = 3.3 ± 1.1). All groups were somewhat confident in study. Even when naps were added to nighttime sleep their prediction (ML = 3.7 ± 1.4, MP = 3.9 ± 1.3, to yield total sleep time, sleep during the study was still EL = 3.2 ± 1.2). There were no statistically significant dif- about an hour less than before. The only significant differences among the groups for either item, determined ference between groups was that the ML group reported by 1-way ANOVAs. There were small correlations be- slightly more sleep during the 4 treatment weeks com- tween expectations and depression ratings, showing a pared with the EL group (Ϸ20 minutes per day more). trend for patients who expected a greater improvement There were no significant differences in sleep between to achieve a greater reduction in their depression (ex- the ML and MP groups.

ARCH GEN PSYCHIATRY/ VOL 55, OCT 1998 885

©1998 American Medical Association. All rights reserved. Downloaded From: https://jamanetwork.com/ on 09/26/2021 30 80 Morning Light Morning Light Morning Placebo Morning Placebo Evening Light Evening Light 25 ∗ 60 ∗ 20

15 40 % Responders 10 24-Item SIGH-SAD Score 20

0 0 Pre-baseline Baseline 1 2 3 4 1 2 3 4 Week Treatment Week Figure 1. Mean Structured Interview Guide for the Hamilton Depression Figure 2. Percent of patients with nearly complete remissions, defined as Rating Scale, Seasonal Affective Disorder Version (SIGH-SAD) depression achieving both a 50% decrease (from baseline) in the 24-item Structured ratings (17 original Hamilton items plus 7 atypical items) for a week previous Interview Guide for the Hamilton Depression Rating Scale, Seasonal Affective to the baseline week, for the baseline week, and for each of the 4 treatment Disorder Version (SIGH-SAD) and a score of 8 or lower. Asterisk indicates weeks. PϽ.05 compared with placebo.

DEPRESSION RATINGS was higher than for MP at week 3 (z = 2.55, PϽ.05) and at week 4 (z = 2.24, PϽ.05). In addition, there were more Figure 1 shows that all 3 treatments greatly reduced de- responders to ML than EL at each treatment week, but pression, but that there was very little difference be- this difference was only significant during treatment week tween the treatments. A repeated-measures ANOVA was 3(z= 2.13, PϽ.05). We also analyzed these data by ap- performed on the difference from baseline scores. The plying even stricter response criteria. Responders had to between-subjects factor was treatment group (ML, EL, meet the same criteria for complete remission as men- MP) and the within-subjects factor was time (treatment tioned in Figure 2, but they had to meet these criteria in weeks 1 through 4). There was a significant effect of time both treatment weeks 3 and 4. By these criteria the re- (F3,279 = 61.15, PϽ.001), but not of treatment group sponse rates for the ML, EL, and MP groups were 55%, (F2,93 = 0.28, P = .75). The interaction of group ϫ time was 28%, and 16%, respectively. Morning light was signifi- also not significant (F6,279 = 1.01, P = .42). Thus, patients cantly better than MP (z = 3.05, PϽ.01) and signifi- improved over time regardless of treatment group. Two cantly better than EL (z = 2.13, PϽ.05). tests, the Mauchly sphericity test and the Box M, both The reason that significant differences between revealed that the assumption of homogeneous vari- groups emerged when calculating percent responders ances was violated. Therefore, it would not be advisable (Figure 2), but not with mean ratings (Figure 1) can be to rely solely on the ANOVAs for statistical significance. understood by considering the distributions of scores We classified patients as responders if their 24- (Figure 3). The scatterplots show that more patients item SIGH-SAD score decreased to 50% of baseline. The met the strict joint criteria for remission in the ML percent responders for the ML, EL, and MP groups were group than in the MP or EL groups. The frequency his- 55%, 56%, and 52% after 3 weeks of treatment, and 67%, tograms further illustrate how the scores were distrib- 75%, and 48% after 4 weeks of treatment. To determine uted differently in the 3 groups. The ML distribution whether these percentages were significantly different at was Poisson-like, with scores clustered at the bottom, each treatment week, log-linear analyses were used to whereas the MP distribution was more normal. In sum- compare the proportion of responders and nonre- mary, although the mean scores were similar among the sponders among the 3 groups (equivalent to 2 ϫ 3 ␹2 3 groups (Ϸ10-12), ML produced the greatest number tests). There were no statistically significant differences of full remissions. at either treatment week 3 or treatment week 4. The results from the Beck depression ratings were Significant differences between treatments finally similar to the SIGH-SAD ratings. There was little differ- emerged when we classified patients as responders by strict ence among the 3 groups in mean weekly ratings. Log- joint criteria designed to identify those with complete or linear analyses on percent responders by strict criteria nearly complete remissions (a decrease in SIGH-SAD of showed significantly higher response rates to ML than 50% and a score Յ8), criteria similar to those used pre- MP. However, there were no significant differences be- viously.21 Figure 2 shows that the percent responders tween ML and EL (Table 3). increased as the weeks progressed, and that ML produced the best response rates. At treatment week 4, the COMMENT percent responders for the ML, EL, and MP groups were 61%, 50%, and 32%, respectively. Log-linear analyses re- Bright light therapy was a better antidepressant for SAD vealed significant differences among treatments; con- than placebo, but the difference did not reach statistical trast parameters showed that the response rate for ML significance until the third week of treatment. A lag of

ARCH GEN PSYCHIATRY/ VOL 55, OCT 1998 886

30 30 Morning Light Morning Placebo Evening Light Weekly Ratings* 20 20 Pre-baseline 24.0 (8.1) 27.0 (8.0) 25.4 (8.1) Baseline 22.0 (9.2) 25.7 (10.7) 23.6 (10.8) 10 10 Treatment week 1 15.0 (7.5) 18.3 (9.0) 18.7 (9.4)

Treatment Weeks 3 and 4 Weeks Treatment Treatment week 2 12.8 (9.7) 15.7 (8.1) 15.2 (9.6) Treatment week 3 9.3 (8.7) 12.5 (7.3) 10.7 (6.5) 0 10 20 30 40 0 10 20 30 40 Treatment week 4 8.2 (8.9) 10.8 (7.9) 9.1 (6.4) Baseline Baseline Percent Responders (Score Յ50% of Baseline and Յ8) C D 40 EL 40 ML MP EL Treatment week 1 18.2 12.9 3.1 Treatment week 2 24.2 19.4 15.6 30 30 Treatment week 3 57.6† 22.6 43.8 Treatment week 4 57.6 35.5 43.8 Both weeks 3 and 4 51.5† 22.6 34.4 20 20 *Weekly ratings are given as mean (SD). 10 10 †Significantly different from morning placebo (PϽ.05). Treatment Weeks 3 and 4 Weeks Treatment

0 10 20 30 40 0 10 0 10 0 10 known that the apparent efficacy of antidepressant treat- Baseline Frequency ments decreases as designs are made more sophisti- Figure 3. A, B, and C, Scatterplots show depression ratings the 24-item cated; eg, with more adequate control treatments and bet- Structured Interview Guide for the Hamilton Depression Rating Scale, ter blinding techniques.10,12,13,33 Seasonal Affective Disorder Version (SIGH-SAD). Each point shows 1 patient’s score during the baseline week and during treatment weeks 3 and 4 It has been suggested that a certain amount of posi- (averaged). Points that fall below the 45° diagonal represent patients who tive expectations or placebo effect may be necessary to improved. Points that fall below the lower diagonal (which ends at 20) release the specific pharmacologic effect of a drug or the represent patients whose scores were reduced by 50% or more. Points that specific effects of other treatments.34-36 This is one rea- fall on or below the horizontal line at 8 and below the lower diagonal represent patients who met the joint criteria of having their scores reduced son we strived to increase the positive expectations of by 50% and to 8 or lower. D, Frequency distributions of the treatment scores our patients. Thus, as expected, there was a larger pla- in the scatter plots. ML indicates morning light; MP, morning placebo; and cebo response rate in the current study than in our pre- EL, evening light. vious study17: 36% vs 22% of patients were responders at least 3 weeks for significant antidepressant effects to with placebo (Յ50% of baseline with 2 weeks of MP). develop is consistent with antidepressant drug stud- Similarly, there was a larger response rate to bright light, ies.22 Thus, it seems likely that the reason our previous 49% vs 34% (with 2 weeks of ML). The increase in re- efficacy study17 and the other light box study with a good sponse rate for light was similar to the increase for pla- placebo control (no light, infrared deception)16 could not cebo. Thus, the increased response rate to light in the cur- demonstrate a difference between light and placebo was rent study can be accounted for by the increased placebo because the treatment time was too short. In both cases, component rather than factors specific to light treat- the light treatment was only 2 weeks. The infrared de- ment, such as the increase in daily dose from 1.0 to 1.5 ception study16 also included head-mounted light vi- hours. Conversely, the low response rate to light in the sors, but there was no difference in antidepressant re- previous study can be attributed to a smaller placebo com- sponse between light and no-light visors. There have been ponent, rather than a deficient dose of light. 4 other visor studies,23-26 all with 1- or 2-week treat- In this study, the largest component of the antide- ments, that did not show a superiority for the presumed pressant response to bright light was the placebo effect, active visor despite large sample sizes. These studies used as in antidepressant drug studies.13 The placebo effect is parallel designs, so that patients only saw their own vi- composed of many factors, including anxiety reduction sor and could not tell if it was a placebo. We believe that associated with receiving a diagnosis and positive these 5 studies16,23-26 are the best efficacy studies of light expectations for treatment. In our study, another factor therapy published so far, because data showed that pa- was the regular sleep schedule, which could have func- tients did not have higher expectations for the pre- tioned as a zeitgeber to stabilize circadian rhythms and sumed active treatment than for the placebo. improve sleep.29,37,38 The restrictions on drinking alco- However, there have been several SAD light box stud- hol and caffeinated beverages also constitute basic sleep ies that found bright light more antidepressant than dim hygiene and may have improved sleep. On the other light, sometimes dramatically better after only 1 or 2 hand, the strict early sleep schedule produced chronic weeks.7,27-31 It is possible that those differences were pri- partial sleep deprivation, which could have exacerbated marily produced by differences in expectation/placebo symptoms, especially fatigue, or conceivably produced effects because these were light box studies, mostly cross- an antidepressant effect.39,40 Thus, there are many non- over designs, in which the bright vs dim comparison was specific factors that could have contributed to the pla- obvious. This explanation is supported by an analysis of cebo effect. The fact that we found only small correla- light box studies showing that responders had higher ex- tions between expectation ratings and antidepressant pectations than nonresponders.32 Furthermore, it is well response makes sense given that patient expectations

ARCH GEN PSYCHIATRY/ VOL 55, OCT 1998 887

©1998 American Medical Association. All rights reserved. Downloaded From: https://jamanetwork.com/ on 09/26/2021 were only one of the factors influencing antidepressant Murakami, Marilyn Perlman, Kay Peterson, Jill M. Rich- response. The important principle for an efficacy study ards, and Kelly J. Sherack. William Scheftner, MD, re- is to make sure that all the nonspecific factors, includ- viewed the laboratory test and physical examination re- ing expectations, are the same in both the active and ports made during the screening process. Research assistants placebo conditions. in the Biological Rhythms Research Laboratory of the De- Because we succeeded in creating a placebo treat- partment of Psychology who saw patients and entered data ment that was similar to light treatment in many ways into the computer were David Anyadike, Linda C. Gallo, (an electrical device that creates an environmental fac- Christina Hardway, Erin K. Hoese, Michael P. Mahoney, tor, 1.5 hours of regular sitting time, the early sleep sched- Paula J. Mitchell, PhD, Trina Moss, Charles Splete, and ule, and so on), and because expectation ratings showed Robert A. Tell. Larry D. Chait, PhD, suggested using nega- no difference between light and placebo, we are fairly con- tive ion generators as the placebo treatment. Ron Rehr of fident that bright light had a specific antidepressant ef- the Machine Shop and Phillip Zervos of the Electronic fect beyond its placebo effect. However, our placebo could Research Facility of Rush-Presbyterian-St Luke’s Medical be criticized for being obviously different from a light box. Center helped design and build the negative ion generators. Our expectation questionnaire, although purposely given Reprints: Charmane I. Eastman, PhD, Biological by research assistants (not by more “formidable” PhDs) Rhythms Research Laboratory, Rush-Presbyterian-St Luke’s and without any pressure for specific responses, still may Medical Center, 1653 W Congress Pkwy, Chicago, IL 60612 not have captured the true feelings of our patients. If they (e-mail: [email protected]). actually had higher expectations for light, that could account for the superior antidepressant effect of light in our REFERENCES study. On the other hand, our design had an advantage over the typical double-blind antidepressant drug stud- 1. Eastman CI. Natural summer and winter sunlight exposure patterns in seasonal ies, in which raters can often tell which patients were given affective disorder. Physiol Behav. 1990;48:611-616. 13,33 drug by the emergence of side effects. Our raters had 2. Cole RJ, Kripke DF, Wisbey J, Mason WJ, Gruen W, Hauri PJ, Juarez S. Sea- no similar clues to guess which patients had light boxes sonal variation in human illumination exposure at two different latitudes. J Biol and which had negative-ion generators. Rhythms. 1995;10:324-334. There has been a controversy about whether ML is 3. Guillemette J, Hebert M, Paquet J, Dumont M. Natural bright light exposure in the summer and winter in subjects with and without complaints of seasonal mood a better antidepressant than EL, with some studies show- variations. Biol Psychiatry. In press. 14,41-44 ing a superiority for ML, but others showing no dif- 4. Lewy AJ, Kern HA, Rosenthal NE, Wehr TA. Bright artificial light treatment of a ference between ML and EL.15,45-48 We found that ML was manic-depressive patient with a seasonal mood cycle. Am J Psychiatry. 1982; better than EL by some measures but not by others; the 139:1496-1498. difference between ML and EL was not very robust. Those 5. Rosenthal NE. Diagnosis and treatment of seasonal affective disorder. JAMA. 1993;270:2717-2720. differences we did find cannot be attributed to differ- 6. Oren DA, Rosenthal NE. Seasonal affective disorders. In: Paykel ES, ed. ences in expectations, because these 2 groups had simi- Handbook of Affective Disorders. New York, NY: Churchill Livingstone; 1992: lar expectations. However, patients receiving EL ob- 551-564. tained slightly less sleep during the treatment weeks than 7. Rosenthal NE, Sack DA, Gillin JC, Lewy AJ, Goodwin FK, Davenport Y, Mueller PS, Newsome DA, Wehr TA. Seasonal affective disorder: a description of the syn- patients receiving ML. Thus, ML could have produced a drome and preliminary findings with light therapy. Arch Gen Psychiatry. 1984; slightly better antidepressant response because it ad- 41:72-80. vances circadian rhythms and can help patients adapt to 8. Jones RA. Self-Fulfilling Prophecies: Social, Psychological, and Physiological Ef- an earlier sleep schedule, whereas EL delays rhythms and fects of Expectancies. New York, NY: John Wiley & Sons; 1977. can make it more difficult to adjust.49 The ML and EL 9. Shapiro AK, Morris LA. The placebo effect in medical and psychological therapies. In: Garfield SL, Bergin AE, eds. Handbook of Psychotherapy and Behavior Change: groups also differed in the restrictions on social activi- An Empirical Analysis. New York, NY: John Wiley & Sons; 1978:369-410. ties. Patients receiving EL had to stay at home (average 10. Ross M, Olson JM. An expectancy-attribution model of the effects of placebos. time 9:00-10:30 PM) to receive their treatment. This in- Psychol Rev. 1981;88:408-437. terfered with many social and family activities and could 11. Lundh LG. Placebo, belief, and health: a cognitive-emotional model. Scand J Psy- have worked against an antidepressant effect. If we had chol. 1987;28:128-143. 12. Eastman CI. What the placebo literature can tell us about light therapy for SAD. included an evening placebo group, we might have been Psychopharmacol Bull. 1990;26:495-504. able to learn more. 13. Greenberg RP, Fisher S. Examining antidepressant effectiveness: findings, am- biguities, and some vexing puzzles. In: Fisher S, Greenberg RP, eds. The Limits Accepted for publication March 24, 1998. of Biological Treatments for Psychological Distress. Hillsdale, NJ: Lawrence A Erlbaum Associates; 1989:1-37. This research was supported by grant MH42768 from 14. Lewy AJ, Sack RL, Miller LS, Hoban TM. Antidepressant and circadian phase- the National Institutes of Mental Health, Rockville, Md. shifting effects of light. Science. 1987;235:352-354. Presented in part at the Werner-Gren Center Interna- 15. Wirz-Justice A, Graw P, Krauchi K, Gisin B, Jochum A, Arendt J, Fisch H-U, tional Symposium on Light and Biological Rhythms in Man, Buddeberg C, Poldinger W. Light therapy in seasonal affective disorder is inde- pendent of time of day or circadian phase. Arch Gen Psychiatry. 1993;50: Stockholm, Sweden, September 12, 1992, and the Eighth An- 929-937. nual Meeting of the Society for Light Treatment and Bio- 16. Levitt AJ, Wesson VA, Joffe RT, Maunder RG, King EF. A controlled comparison logical Rhythms, Washington, DC, June 3, 1996. of light box and head-mounted units in the treatment of seasonal depression. Eva Chernin and Patricia Meaden, PhD, were re- J Clin Psychiatry. 1996;57:105-110. search coordinators in the Department of Psychiatry, Rush- 17. Eastman CI, Lahmeyer HW, Watell LG, Good GD, Young MA. A placebo-control led trial of light treatment for winter depression. J Affect Disord. 1992;26: Presbyterian-St Luke’s Medical Center, Chicago, Ill. Other 211-222. staff in the Department of Psychiatry who helped with re- 18. Williams JBW, Link MJ, Rosenthal NE, Amira L, Terman M. Structured Inter- cruitment, diagnoses, and SIGH-SAD ratings were James view Guide for the Hamilton Depression Rating Scale–Seasonal Affective Disor-

ARCH GEN PSYCHIATRY/ VOL 55, OCT 1998 888

©1998 American Medical Association. All rights reserved. Downloaded From: https://jamanetwork.com/ on 09/26/2021 der Version (SIGH-SAD). New York, NY: NY State Psychiatric Institute; 1992. 34. Uhlenhuth EH, Canter A, Neustadt JO, Payson HE. The symptomatic relief of 19. Beck AT, Ward CH, Mendelson M, Mock J, Erbaugh J. An inventory for measur- anxiety with meprobamate, phenobarbital and placebo. Psychiatry. 1959;115: ing depression. Arch Gen Psychiatry. 1961;4:53-63. 905-910. 20. Bishop YMM, Fienberg SE, Holland PW. Discrete Multivariate Analysis. Cam- 35. Shapiro AK. Placebo effects in medicine, psychotherapy, and psychoanalysis. bridge, Mass: MIT Press; 1975. In: Bergin AE, Garfield SL, eds. Handbook of Psychotherapy and Behavior Change: 21. Terman M, Terman JS, Quitkin FM, McGrath PJ, Stewart JW, Rafferty B. Light An Empirical Analysis. New York, NY: John Wiley & Sons; 1971:439-473. therapy for seasonal affective disorder: a review of efficacy. Neuropsychophar- 36. Grunbaum A. The placebo concept in medicine and psychiatry. Psychol Med. 1986; macology. 1989;2:1-22. 16:19-38. 22. Janicak PG, Davis JM, Preskorn SH, Ayd FJ Jr. Principles and Practice of Psy- 37. Monk TH, Petrie SR, Hayes AJ, Kupfer DJ. Regularity of daily life in relation to chopharmacotherapy. Baltimore, Md: Williams & Wilkins; 1997. personality, age, gender, sleep quality and circadian rhythms. J Sleep Res. 1994; 23. Rosenthal NE, Moul DE, Hellekson CJ, Oren DA, Frank A, Brainard GC, Murray 3:196-205. MG, Wehr TA. A multicenter study of the light visor for seasonal affective dis- 38. Manber R, Bootzin RR, Acebo C, Carskadon MA. The effect of regularizing sleep- order: no difference in efficacy found between two different intensities. Neuro- wake schedules on daytime sleepiness. Sleep. 1996;19:432-441. psychopharmacology. 1993;8:151-160. 39. Schilgen B, Tolle R. Partial sleep deprivation as therapy for depression. Arch Gen 24. Joffe RT, Moul DE, Lam RW, Levitt AJ, Teicher MH, Lebegue B, Oren DA, Buchanan Psychiatry. 1980;37:267-271. A, Glod CA, Murray MG, Brown J, Schwartz P. Light visor treatment for sea- 40. Vogel GW, Vogel F, McAbee RS, Thurmond AJ. Improvement of depression by sonal affective disorder: a multicenter study. Psychiatry Res. 1993;46:29-39. REM sleep deprivation. Arch Gen Psychiatry. 1980;37:247-253. 25. Levitt AJ, Joffe RT, King E. Dim versus bright red (light-emitting diode) light in 41. Sack RL, Lewy AJ, White DM, Singer CM, Fireman MJ, Vandiver R. Morning vs the treatment of seasonal affective disorder. Acta Psychiatr Scand. 1994;89:341- evening light treatment for winter depression. Arch Gen Psychiatry. 1990;47: 345. 343-351. 26. Teicher MH, Glod CA, Oren DA, Schwartz PJ, Luetke C, Brown C, Rosenthal NE. 42. Avery DH, Khan A, Dager SR, Cox GB, Dunner DL. Bright light treatment of win- The phototherapy light visor: more to it than meets the eye. Am J Psychiatry. ter depression: morning versus evening light. Acta Psychiatr Scand. 1990;82: 1995;152:1197-1202. 335-338. 27. Rosenthal NE, Sack DA, Carpenter CJ, Parry BL, Mendelson WB, Wehr TA. An- 43. Avery DH, Khan A, Dager SR, Cohen S, Cox GB, Dunner DL. Morning or evening tidepressant effects of light in seasonal affective disorder. Am J Psychiatry. 1985; bright light treatment of winter depression? the significance of hypersomnia. Biol 142:163-170. Psychiatry. 1991;29:117-126. 28. James SP, Wehr TA, Sack DA, Parry BL, Rosenthal NE. Treatment of seasonal 44. Terman JS, Terman M, Schlager D, Rafferty B, Rosofsky M, Link MJ, Gallin PF, affective disorder with light in the evening. Br J Psychiatry. 1985;147:424-428. Quitkin FM. Efficacy of brief, intense light exposure for treatment of winter de- 29. Kohler WK, Pelzer A, Schmidt KP, Carella AB, Pflug B. Bright light and dim light pression. Psychopharmacol Bull. 1990;26:3-11. in the therapy of depression: effects on the circadian system and clinical results. 45. Hellekson CJ, Kline JA, Rosenthal NE. Phototherapy for seasonal affective dis- Pharmacopsychiatry. 1989;22:202. order in Alaska. Am J Psychiatry. 1986;143:1035-1037. 30. Winton F, Corn T, Huson LW, Franey C, Arendt J, Checkley SA. Effects of light 46. Meesters Y, Jansen JHC, Lambers PA, Bouhuys AL, Beersma DGM, van den Hoof- treatment upon mood and melatonin in patients with seasonal affective disor- dakker RH. Morning and evening light treatment of seasonal affective disorder: der. Psychol Med. 1989; 19:585-590. response, relapse and prediction. J Affect Disord. 1993;28:165-177. 31. Magnusson A, Kristbjarnarson H. Treatment of seasonal affective disorder with 47. Meesters Y, Jansen JHC, Beersma DGM, Bouhuys AL, van den Hoofdakker RH. high-intensity light: a phototherapy study with an Icelandic group of patients. Light therapy for seasonal affective disorder: the effects of timing. Br J Psychia- J Affect Disord. 1991;21:141-147. try. 1995;166:607-612. 32. Terman M, Amira L, Terman JS, Ross DC. Predictors of response and nonre- 48. Lafer B, Sachs GS, Labbate I-A, Thibault A, Rosenbaum JF. Phototherapy for sea- ponse to light treatment for winter depression. Am J Psychiatry. 1996;153: sonal affective disorder: a blind comparison of three different schedules. Am J 1423-1429. Psychiatry. 1994;151:1081-1083. 33. Fisher S, Greenberg RP. How sound is the double-blind design for evaluating 49. Eastman CI. Circadian rhythms and bright light: recommendations for shift work. psychotropic drugs? J Nerv Ment Disord. 1993;181:345-350. Work Stress. 1990;4:245-260.

ARCH GEN PSYCHIATRY/ VOL 55, OCT 1998 889